Search NASA⌕ Search

SEARCH · Search NASA

Results for “Kernel methods”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Sparse-Stochastic Fragmented Exchange for Large-Scale Hybrid Time-Dependent Density Functional Theory Calculations

Here we extend our recently developed sparse-stochastic fragmented exchange formalism for ground-state near-gap hybrid DFT to calculate absorption spectra within linear-response time-dependent generalized Kohn-Sham DFT (LR-GKS-TDDFT) for systems consisting of thousands of valence electrons within a grid-based/plane-wave representation. A mixed deterministic/fragmented-stochastic compression of the exchange kernel, here using long-range explicit exchange functionals, provides an efficient method for accurate optical spectra. Both real-time propagation as well as frequency-resolved Casida-equation-type approaches for spectra are presented, and the method is applied to large molecular dyes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2014

The rigorous stochastic matrix multiplication scheme for the calculations of reduced equilibrium density matrices of open multilevel quantum systems

Understanding the roles of the temporary and spatial structures of quantum functional noise in open multilevel quantum molecular systems attracts a lot of theoretical interests. I want to establish a rigorous and general framework for functional quantum noises from the constructive and computational perspectives, i.e., how to generate the random trajectories to reproduce the kernel and path ordering of the influence functional with effective Monte Carlo methods for arbitrary spectral densities. This construction approach aims to unify the existing stochastic models to rigorously describe the temporary and spatial structure of Gaussian quantum noises. In this paper, I review the Euclidean imaginary time influence functional and propose the stochastic matrix multiplication scheme to calculate reduced equilibrium density matrices (REDM). In addition, I review and discuss the Feynman-Vernon influence functional according to the Gaussian quadratic integral, particularly its imaginary part which is critical to the rigorous description of the quantum detailed balance. As a result, I establish the conditions under which the influence functional can be interpreted as the average of exponential functional operator over real-valued Gaussian processes for open multilevel quantum systems. I also show the difference between the local and nonlocal phonons within this framework. With the stochastic matrix multiplication scheme, I compare the normalized REDM with the Boltzmann equilibrium distribution for open multilevel quantum systems.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2019

Characteristics of Fuel Articles for Irradiation Testing in INL-TREAT and MIT-NRL

To obtain valuable data that will help in the design and commissioning of the Transformational Challenge Reactor, a series of graded, increasingly complex irradiation tests are planned that will demonstrate the feasibility of advanced manufactured reactor core components. In the first irradiation test, conventionally- fabricated uranium nitride tristructural isotropic (UN TRISO) fuel kernels will be embedded into compacts that are fabricated using advanced manufacturing methods. These compacts are to be irradiated at Massachusetts Institute of Technology’s Nuclear Reactor Laboratory and Idaho National Laboratory’s Transient Reactor Test Facility. This report presents the characteristics of the proposed samples and experiments.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
NASA NTRS1986

Numerical solution of random singular integral equation appearing in crack problems

The solution of several elasticity problems, and particularly crack problems, can be reduced to the solution of one-dimensional singular integral equations with a Cauchy-type kernel or to a system of uncoupled singular integral equations. Here a method for the numerical solution of random singular integral equations of Cauchy type is presented. The solution technique involves a Chebyshev series approximation, the coefficients of which are the solutions of a system of random linear equations. This method is applied to the problem of periodic array of straight cracks inside an infinite isotropic elastic medium and subjected to a nonuniform pressure distribution along the crack edges. The statistical properties of the random solution are evaluated numerically, and the random solution is used to determine the values of the stress-intensity factors at the crack tips. The error, expressed as the difference between the mean of the random solution and the deterministic solution, is established. Values of stress-intensity factors at the crack tip for different random input functions are presented.

Sambandham, M.↗
DOE OSTI2016

Cylindrical Kernel Density Estimators For Monte Carlo Neutron Transport Reactor Physics Problems

Kernel Density Estimators (KDEs) have been investigated as an alternative to histogram tallies in Monte Carlo radiation transport simulations for capturing spatially-resolved quantities such as scalar flux and reaction rate densities. KDEs are non-parametric estimators that show potential for producing smooth estimates of quantities of interest with reduced variance compared to traditional histogram tallies. Furthermore, KDEs obtain estimates of the underlying distribution at user-defined points without requiring an underlying mesh, thus making KDEs an attractive potential alternative to histogram tallies. Recent work has introduced the Mean Free Path based KDE (MFP KDE) in order to capture distributions with large curvature at material interfaces, e.g. absorption reaction rates in UO{sub 2} fuel in light water reactors. However, the multivariate MFP KDE produces inaccuracies in the form of localized spikes in problems with material cross sections that contain large resonances. Since the MFP KDE works well in one dimension, it is hypothesized that a different MFP KDE formulation where the normalization coefficient only contains one factor of the cross section would be able to capture these distributions with large curvature without producing spikes in the estimate. This can be accomplished by using a cylindrical KDE whose 2-D support area is wedge-shaped and applying the MFP KDE to the radial component of the cylindrical KDE. These cylindrical KDEs use 2-D (r,θ) cylindrical coordinates with no dependence in axial dimension z. This paper derives the cylindrical KDE and the cylindrical MFP KDE and tests them on a reactor physics pin-cell benchmark problem, producing accurate estimates of flux and reaction rate densities with differences less than 0.5% when compared to a reference histogram. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2017

Spherical Kernel Density Estimators For Monte Carlo Radiation Transport Problems

Recently, there has been interest in creating alternatives to histograms (mesh tallies) for estimating spatially-distributed quantities such as scalar flux and reaction rate densities. Kernel Density Estimators (KDEs) are non-parametric estimators that obtain estimates of the underlying distribution at user-defined points without having to specify an underlying mesh, and the uncertainty at those points is independent of the number of tally points used in the simulation. This has led to the investigation of KDEs as an attractive potential alternative to histogram tallies in Monte Carlo radiation transport simulations for reactor physics problems and shielding applications. A KDE was recently created specifically for cylindrical geometries that is capable of producing smooth and accurate estimates of flux and reaction rate densities in reactor physics problems. This work extends the cylindrical KDE and cylindrical Mean Free Path (MFP) KDE to spherical geometries and verifies the spherical KDE on bare and reflected critical benchmark problems.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2012

Improved Moderator Performance Calculations at SNS

The Spallation Neutron Source (SNS) at Oak Ridge National Laboratory reached a nominal operating power of 1 MW in September, 2009. Even at ~2/3 of its designed capacity, SNS already becomes the brightest pulsed neutron source in the world, and a premier facility for material and life science research. To provide accurate target performance estimation and to meet future development needs, we had developed an as-built model for the operating target system of SNS. Based on this asbuilt model, the moderator performance at SNS was recalculated taking advantage of the improved physics and tallying method in the Monte Carlo simulation code, MCNPX, and most recent scattering kernels. The neutron flux intensity spectra, emission time distributions of neutron pulses and moderator brightness profiles regarding the SNS moderator performance are presented and discussed. Compared to the previous results, the current calculations show ~10-20% lower in flux with improved emission time distributions of neutron pulses. The difference between the two data sets is analyzed, as well as other factors impacting the moderator performance at SNS. The current calculation results will serve as source terms for the instrument performance simulations at SNS.

Lu, Wei↗
DOE OSTI2014

Motion as perturbation. II. Development of the method for dosimetric analysis of motion effects with fixed-gantry IMRT

Purpose: In this work, the feasibility of implementing a motion-perturbation approach to accurately estimate volumetric dose in the presence of organ motion—previously demonstrated for VMAT-–is studied for static gantry IMRT. The method's accuracy is improved for the voxels that have very low planned dose but acquire appreciable dose due to motion. The study describes the modified algorithm and its experimental validation and provides an example of a clinical application. Methods: A contoured region-of-interest is propagated according to the predefined motion kernel throughout time-resolved 4D phantom dose grids. This timed series of 3D dose grids is produced by the measurement-guided dose reconstruction algorithm, based on an irradiation of a staticARCCHECK (AC) helical dosimeter array (Sun Nuclear Corp., Melbourne, FL). Each moving voxel collects dose over the dynamic simulation. The difference in dose-to-moving voxel vs dose-to-static voxel in-phantom forms the basis of a motion perturbation correction that is applied to the corresponding voxel in the patient dataset. A new method to synchronize the accelerator and dosimeter clocks, applicable to fixed-gantry IMRT, was developed. Refinements to the algorithm account for the excursion of low dose voxels into high dose regions, causing appreciable dose increase due to motion (LDVE correction). For experimental validation, four plans using TG-119 structure sets and objectives were produced using segmented IMRT direct machine parameters optimization in Pinnacle treatment planning system (v. 9.6, Philips Radiation Oncology Systems, Fitchburg, WI). All beams were delivered with the gantry angle of 0°. Each beam was delivered three times: (1) to the static AC centered on the room lasers; (2) to a static phantom containing a MAPCHECK2 (MC2) planar diode array dosimeter (Sun Nuclear); and (3) to the moving MC2 phantom. The motion trajectory was an ellipse in the IEC XY plane, with 3 and 1.5 cm axes. The period was 5 s, with the resulting average motion speed of 1.45 cm/s. The motion-perturbed high resolution (2 mm voxel) volumetric dose grids on the MC2 phantom were generated for each beam. From each grid, a coronal dose plane at the detector level was extracted and compared to the corresponding moving MC2 measurement, using gamma analysis with both global (G) and local (L) dose-error normalization. Results: Using the TG-119 criteria of (3%G/3 mm), per beam average gamma analysis passing rates exceeded 95% in all cases. No individual beam had a passing rate below 91%. LDVE correction eliminated systematic disagreement patterns at the beams’ aperture edges. In a representative example, application of LDVE correction improved (2%L/2 mm) gamma analysis passing rate for an IMRT beam from 74% to 98%. Conclusions: The effect of motion on the moving region-of-interest IMRT dose can be estimated with a standard, static phantom QA measurement, provided the motion characteristics are independently known from 4D CT or otherwise. The motion-perturbed absolute dose estimates were validated by the direct planar diode array measurements, and were found to reliably agree with them in a homogeneous phantom.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2024

Forward variable selection enables fast and accurate dynamic system identification with Karhunen-Loève decomposed Gaussian processes

A promising approach for scalable Gaussian processes (GPs) is the Karhunen-Loève (KL) decomposition, in which the GP kernel is represented by a set of basis functions which are the eigenfunctions of the kernel operator. Such decomposed kernels have the potential to be very fast, and do not depend on the selection of a reduced set of inducing points. However KL decompositions lead to high dimensionality, and variable selection thus becomes paramount. This paper reports a new method of forward variable selection, enabled by the ordered nature of the basis functions in the KL expansion of the Bayesian Smoothing Spline ANOVA kernel (BSS-ANOVA), coupled with fast Gibbs sampling in a fully Bayesian approach. It quickly and effectively limits the number of terms, yielding a method with competitive accuracies, training and inference times for tabular datasets of low feature set dimensionality. Theoretical computational complexities are O ( N P 2 ) in training and O ( P ) per point in inference, where N is the number of instances and P the number of expansion terms. The inference speed and accuracy makes the method especially useful for dynamic systems identification, by modeling the dynamics in the tangent space as a static problem, then integrating the learned dynamics using a high-order scheme. The methods are demonstrated on two dynamic datasets: a ‘Susceptible, Infected, Recovered’ (SIR) toy problem, along with the experimental ‘Cascaded Tanks’ benchmark dataset. Comparisons on the static prediction of time derivatives are made with a random forest (RF), a residual neural network (ResNet), and the Orthogonal Additive Kernel (OAK) inducing points scalable GP, while for the timeseries prediction comparisons are made with LSTM and GRU recurrent neural networks (RNNs) along with the SINDy package.

Hayes, Kyle↗
DOE OSTI2018

Modeling error PDF optimization based wavelet neural network modeling of dynamic system and its application in blast furnace ironmaking

In general, the modeling errors of dynamic system model are a set of random variables. The traditional performance index of modeling such as means square error (MSE) and root means square error (RMSE) can not fully express the connotation of modeling errors with stochastic characteristics both in the dimension of time domain and space domain. Therefore, the probability density function (PDF) is introduced to completely describe the modeling errors in both time scales and space scales. Based on it, a novel wavelet neural network (WNN) modeling method is proposed by minimizing the two-dimensional (2D) PDF shaping of modeling errors. First, the modeling error PDF by the tradional WNN is estimated using data-driven kernel density estimation (KDE) technique. Then, the quadratic sum of 2D deviation between the modeling error PDF and the target PDF is utilized as performance index to optimize the WNN model parameters by gradient descent method. Since the WNN has strong nonlinear approximation and adaptive capability, and all the parameters are well optimized by the proposed method, the developed WNN model can make the modeling error PDF track the target PDF, eventually. Simulation example and application in a blast furnace ironmaking process show that the proposed method has a higher modeling precision and better generalization ability compared with the conventional WNN modeling based on MSE criteria. Furthermore, the proposed method has more desirable estimation for modeling error PDF that approximates to a Gaussian distribution whose shape is high and narrow.

Modelling↗
DOE OSTI2014

TU-C-BRE-10: A Streamlined Approach to EPID Transit Dosimetry

Purpose: To investigate the feasibility of a simple and efficient transit dosimetry method using the electronic portal imaging device (EPID) for dose delivery error detection and prevention. Methods: In the proposed method, 2D reference transit images are generated for comparison with online images acquired during treatment. Reference transit images are generated by convolving through-air EPID measurements of each field with pixel-specific kernels selected from a library of pre-calculated Monte Carlo pencil kernels of varying radiological thickness. The kernel used for each pixel is selected based on the calculated radiological thickness of the patient along a line joining the pixel and the virtual source. The accuracy of the technique was evaluated in flat homogeneous and heterogeneous plastic water phantoms, a heterogeneous cylindrical phantom, and an anthropomorphic head phantom. Gamma criteria of 3%/3 mm was used to quantify the accuracy of the technique for the various cases. Results: An average of 99.9% and 99.7% of the points in the comparison between the measured and predicted images passed a 3%/3mm gamma for the homogeneous and heterogeneous plastic water phantoms, respectively. 97.1% of the points passed for the analysis of the heterogeneous cylindrical phantom. For the anthropomorphic head phantom, an average of 97.8% of points passed the 3%/3mm gamma criteria for all field sizes. Failures were observed primarily in areas of drastic thickness or material changes and at the edges of the fields. Conclusion: The data suggest that the proposed transit dosimetry method is a feasible approach to in vivo dose monitoring. Future research efforts could include implementation for more complex fields and sensitivity testing of the method to setup errors and changes in anatomy. Oncology Data Systems provided partial funding support but did not participate in the collection or analysis of data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

A two-level GPU-accelerated incomplete LU preconditioner for general sparse linear systems

This paper presents a parallel preconditioning approach based on incomplete LU (ILU) factorizations in the framework of Domain Decomposition (DD) for general sparse linear systems. We focus on distributed memory parallel architectures, specifically, those that are equipped with graphic processing units (GPUs). In addition to block-Jacobi, we present general purpose two-level ILU Schur complement-based approaches, where different strategies are presented to solve the coarse-level reduced system. These strategies are combined with modified ILU methods in the construction of the coarse-level operator, in order to effectively remove smooth errors by targeting an algebraically smooth vector. We leverage available GPU-based sparse matrix kernels to accelerate the setup and the solve phases of the proposed ILU preconditioner. We evaluate the efficiency of the proposed methods as a smoother for algebraic multigrid (AMG) and as a preconditioner for Krylov subspace methods on challenging anisotropic diffusion problems and a collection of general sparse matrices.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2012

AGR-1 Irradiated Compact 6-1-1 PIE Report: Evaluation of As-Irradiated Fuel Performance Using Leach Burn Leach, IMGA, Materialography, and X-ray Tomography

The AGR-1 irradiation test train consisted of six independently monitored capsules. All capsules contained twelve compacts in three stacks of four compacts each [Pope 2010]. Compact 6-1-1 was located in capsule 6, at the top of the test train. The compact was in stack 1 (one of the two stacks closest to the core) in position 1 (at the bottom of the 4 compact stack). Compact 6-1-1 was identified in the fabrication data package as LEU01-46T-Z32 [Hunn et al. 2006]. Details of the compact fabrication and measured properties can be found in that data package. The compact lot was made using TRISO-coated particles designated as AGR-1 Baseline fuel. These particles contained 19.7% enriched mixed uranium oxideuranium carbide (UCO) kernels with a O/U atomic ratio of ~1.36 and a C/U atomic ratio of ~0.33 [Pope 2010]. Details of the coated particle fabrication and measured properties can be found in the data package for the AGR-1 Baseline particles [Hunn and Lowden 2006]. The compact was irradiated at the INL Advanced Test Reactor (ATR) to an average calculated burn-up of 15.1% fissions per initial metal atom (FIMA) and an average calculated fast fluence of 3.0 x 10 25 n/m 2 [Sterbentz 2011] at a calculated volume averaged and time averaged temperature of approximately 1111°C [Hawkes 2012]. Capsules were disassembled and preliminary compact and graphite holder inspection was performed [Demkowicz et al. 2011]. Compact 6-1-1 was shipped to ORNL and post-irradiation examination (PIE) was performed according to examination plan PLN-3624 [Demkowicz 2010-2]. All the examination plan objectives listed below were completed except item 6, which was deleted due to limited funding. 1. Determine the inventory of fission products retained in the compact matrix, but not contained by the SiC and determine the number of exposed kernels or particles with defective SiC layers using the deconsolidation-leach-burn-leach method. 2. Gamma count all particles to quantify 137Cs inventory for comparison with calculated values and gamma spectrometric measurements made on the whole compact. Also, attempt to quantify the inventory of any other fission products for which gamma emissions are observed. 3. Characterize each particle based on the 137 Cs/ 144 Ce ratio to identify particles with low Cs retention that may be the result of a failed SiC coating. 4. Gamma count a subset of particles to determine the 110m Ag inventory for comparison with calculated values and gamma spectrometric measurements made on the whole compact. Also, attempt to quantify the inventory of other fission products observed. 5. Microscopically inspect selected particles to determine the condition of the kernels and coatings. This could include kernel migration, coating microstructures and defects, fission product attack of the SiC layer, and fission product migration in the coating layers. 6. Perform high temperature fission product release tests on a subset of particles in order to calibrate the collection efficiency of the Core Conduction Cooldown Test Facility (CCCTF) deposition cup for selected fission products. Gamma count particles before and after the heating test to quantify the released fission product fractions. 7. Archive remaining particles for later use. Compact 6-1-1 LBL began with compact deconsolidation in the Building 3525 hot cell, followed by two nitric acid leaches of the particles and matrix residue using a Soxhlet extraction apparatus. The matrix residue was then burned and the ash was leached twice with boiling nitric acid. The deconsolidated particles underwent an IMGA analysis to determine gamma-emitting isotope inventories for individual particles; the most conspicuous elements with reduced inventories, indicating fission product release, were silver and europium. During the IMGA analysis, particles with low silver were identified and a subset of these was set aside for materialography and x-ray. After the completion of the IMGA study, about 10% of the particles (421) were set aside as an archive and the remainder burned back and leached twice in the Soxhlet apparatus. Figure 1 shows a flow chart detailing the PIE of Compact 6-1-1.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2016

SU-G-TeP3-13: The Role of Nanoscale Energy Deposition in the Development of Gold Nanoparticle-Enhanced Radiotherapy

Purpose: Gold nanoparticles (GNPs) can enhance radiotherapy effects. The high photoelectric cross section of gold relative to tissue, particularly at lower energies, leads to localized dose enhancement. However in a clinical context, photon energies must also be sufficient to reach a target volume at a given depth. These properties must be balanced to optimize such a therapy. Given that nanoscale energy deposition patterns around GNPs play a role in determining biological outcomes, in this work we seek to establish their role in this optimization process. Methods: The PENELOPE Monte Carlo code was used to generate spherical dose deposition kernels in 1000 nm diameter spheres around 50 nm diameter GNPs in response to monoenergetic photons incident on the GNP. Induced “lesions” were estimated by either a local effect model (LEM) or a mean dose model (MDM). The ratio of these estimates was examined for a range of photon energies (10 keV to 2 MeV), for three sets of linear-quadratic parameters. Results: The models produce distinct differences in expected lesion values, the lower the alpha-beta ratio, the greater the difference. The ratio of expected lesion values remained constant within 5% for energies of 40 keV and above across all parameter sets and rose to a difference of 35% for lower energies only for the lowest alpha-beta ratio. Conclusion: Consistent with other work, these calculations suggest nanoscale energy deposition patterns matter in predicting biological response to GNP-enhanced radiotherapy. However the ratio of expected lesions between the different models is largely independent of energy, indicating that GNP-enhanced radiotherapy scenarios can be optimized in photon energy without consideration of the nanoscale patterns. Special attention may be warranted for energies of 20 keV or below and low alpha-beta ratios.

60 APPLIED LIFE SCIENCES↗
NASA NTRSDate not supplied

Performance Optimization Methods for a Memory-Bound, Unstructured-Grid CFD Application on Massively Parallel GPU Platforms

Computational performance of the FUN3D unstructured-grid computational fluid dynamics (CFD) application on massively parallel GPU environments is memory-bound and highly dependent upon efficient reads from and atomic updates to the irregular cell-, edge-, and node-based data structures. In this talk, we present recent efforts into optimizing select performance-critical kernels on NVIDIA Tesla V100 and A100 GPUs and AMD CDNA MI100 GPUs. A novel use of L2 cache residency controls and asynchronous loads into on-chip shared memory are explored on the A100 GPU for the sparse iterative solver, which is dominated by mixed-precision, sparse matrix vector multiplication. Demonstrations show that these methods improve global memory bandwidth utilization by 13.5% on the A100 GPU. Several techniques are also presented that use registers and/or shared memory to facilitate array transposition and aggregation which combine to reduce the frequency and increase the cache efficiency of floating-point atomic updates to the irregular data structures. These methods are demonstrated to improve the kernel throughput by nearly 500% on select kernels on the AMD MI100 over atomic updates directly to global memory. Overall, both V100 and A100 GPUs outperformed the MI100 GPU on kernels dominated by double-precision atomic updates; however, the techniques demonstrated here reduced the performance gap and improved the MI100 performance.

GPU CPU unstructured CFD memory↗
DOE OSTI2015

System and method for automated object detection in an image

A contour/shape detection model may use relatively simple and efficient kernels to detect target edges in an object within an image or video. A co-occurrence probability may be calculated for two or more edge features in an image or video using an object definition. Edge features may be differentiated between in response to measured contextual support, and prominent edge features may be extracted based on the measured contextual support. The object may then be identified based on the extracted prominent edge features.

97 MATHEMATICS AND COMPUTING↗